Thanks to Kit Baum, recent author of "An Introduction to Stata
Programming," an update to outreg2 is available from ssc.
stats( ) option has been expanded to include the following sub-options:
coef se tstat pval ci ci_low ci_high aster beta N sum_w mean Var sd
sum min max skewness kurtosis p1 p5 p10 p25 p50 p75 p90 p95 p99 corr
pwcorr spearman str( ) cmd( : )
I actually don't like to add functionalities that are covered by other
user-written programs but it looks like the damage has been already
done. I hope this is clear. Adding a new functionality is not that
difficult. All I have to do have it dumped into a table. This feature
has been made into a new option so that users can do it themselves
like this:
* insert correlation under coefficient:
sysuse auto, clear
reg rep78 headroom length turn gear_ratio
outreg2 using myfile, replace stats(coef cmd( r(rho):corr ) )
This has been hard-coded so that it looks like this:
outreg2 using myfile, replace stats(coef corr)
Basically you can insert any macros produced by r-class command. Or
you can write your own program. This means you can pretty much define
and insert whatever you want, i.e. covariance, variable labels,
correlation ratios, one-sided t-statistics, 5th momentum, etc.
For example, coefficient of variation can be defined and inserted like this:
* r-class program for coefficient of variation
prog define coefvar, rclass
syntax varlist(max=2) [if]
gettoken dep indep : varlist
qui sum `indep' `if'
local variation=`r(sd)'/`r(mean)'*100
ret scalar variation=`variation'
end
* test it
sysuse auto, clear
reg rep78 headroom length turn gear_ratio
coefvar rep78 headroom if e(sample)
ret list
* run it
reg rep78 headroom length turn gear_ratio
outreg2 using myfile, replace stats(sd cmd(r(variation): coefvar))
If this were to be hard-coded into outreg2, the r-class program would
be pasted into the bottom of the ado file and the name added to the
list of valid names. That's it.
If someone happened to write a program for producing index or
statistics of interest like this and thought it would be useful to
others, I will be more than happy to put it into the future version of
outreg2 as a sub-option and acknowledge their contribution in the help
file. You don't have to write one specifically for this purpose, but
if you had something that you thought should be a standard part of
table-making commands, then this would be one way to do that.
I think I previously mentioned outreg2 is byable, as in -by:- prefix.
It will also produce a standard table of summary statistics. And a
limited functionality of n-way cross-tabulations. In hindsight I
probably should not have sliced in codes that was written several
years ago for something else but that's how these guys appear in
outreg2. I didn't have the time to clean up the codes or the syntax
and so they are.
outreg2 also has a limited capacity for [in] [if] [weight]. They are
currently left undocumented. They don't work for cross-tabulation. If
you find an example where it works funny (strange funny) then you can
send it to me and I will take a look at it.
What outreg2 is good at is permutation of wide variety of table. With
a little imagination you can produce pretty much whatever you want.
Some examples are posted here.
http://www.stata.com/statalist/archive/2009-11/msg00587.html
http://www.stata.com/statalist/archive/2009-11/msg00619.html
Roy
P.S. I just realized there is an obscure bug. If no regression has
been run and outreg2 is invoked with something in the varlist, then it
will produce coefficients associated with the varlist. This should be
fixed the next time around but I thought I would mention it here.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
